Hello,
On 1/13/2015 11:53 PM, Ivan Gerasimov wrote:
Hi!
A useful change!
I agree :-) We haven't publicized the change very much yet, but it is
part of JEP 213, Milling Project Coin,
(http://openjdk.java.net/jeps/213) and went into JDK 9's javac late in 2014.
Might it also make sense to allow anonymous variables in the
try-with-resource statement?
So that something like `try (() -> System.out.println("closed")) {}`
would work...
I don't think that is necessary. In JDK 7, we started out allowing a
general AutoCloseable expression in a try-with-resources statement and
that proved problematic. Supporting a final / effectively final variable
seems to be the right balance. The rationale is written up in the JSR
334 materials.
Would it make sense to also patch
jdk/src/java.base/windows/classes/java/io/FileDescriptor.java in a
similar manner, as it appears to contain the exactly same code?
Good catch.
Thanks,
-Joe
Sincerely yours,
Ivan
On 14.01.2015 5:46, joe darcy wrote:
Hello,
Earlier in JDK 9, a language change was made so that if a resource
for a try-with-resources statement is already final or effectively
final, a new resource variable does *not* need to be declared to
manage the variable.
The java.base module should take advantage of this language feature.
I've run an experimental checker for such locations over the base
module and found one candidate where the new language feature can be
used; please review this patch for it:
diff -r d873f6a7d16b
src/java.base/unix/classes/java/io/FileDescriptor.java
--- a/src/java.base/unix/classes/java/io/FileDescriptor.java Tue Jan
13 14:33:54 2015 -0800
+++ b/src/java.base/unix/classes/java/io/FileDescriptor.java Tue Jan
13 18:40:54 2015 -0800
@@ -214,7 +214,7 @@
if (!closed) {
closed = true;
IOException ioe = null;
- try (Closeable c = releaser) {
+ try (releaser) {
if (otherParents != null) {
for (Closeable referent : otherParents) {
try {
The build succeeds with this change and the java.io.FileDescriptor
regression tests pass.
Thanks,
-Joe
